Governments need to revamp their procurement processes to effectively acquire artificial intelligence (AI) technologies. To succeed in AI procurement, governments should contextualize data, build high-quality datasets, focus on continuous improvement, include ongoing updates and maintenance in contracts, rigorously evaluate AI models, and support international collaborative frameworks for AI development.
- Contextualize data: Governments must curate specialized, high-quality datasets that comply with security and privacy regulations and adhere to machine learning standards to make AI and automation achievable.
- Perpetually improve: Traditional contracting models are ill-suited for the dynamic nature of AI technology, and contracts should incorporate ongoing updates and maintenance to reflect the iterative nature of AI.
- Rigorously evaluate: Governments should invest in evaluation expertise to scrutinize AI model suitability for mission objectives, data privacy, security norms, and ethical constraints.
- Track and prioritize: Smaller businesses should be supported in scaling their AI solutions within governmental procurement, and programs like Small Business Innovation Research (SBIR) should track and prioritize projects transitioned to programs of record.
- Support international frameworks: Governments should embrace international collaborative frameworks to address the multi-faceted impact of AI, including cooperative security and ethics, for successful AI procurement and utilization.